Applied routing problem for a fleet of delivery drones using a modified parallel genetic algorithm

Abstract

More and more experts agree that in the near future, most freight traffic will be carried out using automated systems, and of them drone delivery is considered to be the most promising. Drone delivery would benefit by independence from the limitations of transport infrastructure and road conditions and would ensure cargo delivery with rapid turnaround times, as well as a significant reduction of environmental impact. The technical capabilities of unmanned aerial vehicles improve year by year, so the task of coordinating drones and effectively planning routes is relevant and in great demand. The development of such technologies will help reduce transportation costs and improve customer service through faster delivery. This article discusses the applied routing problem for a fleet of drones with limited load capacity for the delivery of heterogeneous goods with the possibility of loading in multiple warehouses from an international optimization competition. The solution includes new approach based on a mixed dimensional parallel genetic algorithm (MDPGA) for finding rational routes for delivering goods to various customers and an assignment problem to reduce the dimension depending on the number of warehouses.

Deep neural network based resource allocation in D2D wireless networks

“…After the supervised learning, a brandnew test dataset evaluates the online power allocation in time-varying scenes. The proposed algorithmic technique is implemented using open-source DL framework in PyTorch [26].…”